Photon counting X-ray CT device and method

Abstract

To achieve efficient calibration of a detector response in a photon counting detector having a plurality of groups composed of a plurality of sub-pixels.SOLUTION: A photon counting type X-ray CT apparatus includes a photon counting detector and a processing unit. The photon counting detector includes a plurality of combined pixels composed of a plurality of sub-pixels. The processing unit acquires calibration data on the basis of the sub-pixels, generates calibration data combined according to the spatial resolution on the basis of the above calibration data, and calibrates a detector response of the photon counting detector according to the spatial resolution on the basis of the combined calibration data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The embodiments disclosed herein and in the drawings relate to photon-counting X-ray CT devices and methods. [Background technology]

[0002] Computed tomography (CT) systems and methods are commonly used in medical imaging and diagnosis. CT systems typically create projection images through a subject's body at a series of projection angles. A radiation source, such as an x-ray tube, irradiates the subject's body, producing projection images at various angles. Images of the subject's body can be reconstructed from the projection images.

[0003] Traditionally, energy integrating detectors (EIDs) and / or photon counting detectors (PCDs) are used to measure CT projection data. PCDs offer many advantages, including the ability to perform spectral CT. In this case, PCDs resolve incident x-ray counts into spectral components called energy bins, which collectively span the energy spectrum of the x-ray beam. Unlike non-spectral CT, spectral CT generates information as a function of x-ray energy due to different materials that exhibit different x-ray attenuation at different energies. These differences allow the spectrally resolved projection data to be used to resolve (discriminate) the subject into different reference materials. For example, two constituent materials in material decomposition are bone and water.

[0004] Despite the fast response time of PCDs, at the high x-ray flux rates typical of clinical x-ray imaging, multiple x-ray detection events may occur on a single detector within the detector's time response—a phenomenon known as pile-up. Left uncorrected, the pile-up effect can distort the PCD energy response and degrade images reconstructed from the PCD. Once these effects are corrected, spectral CT offers many advantages over conventional CT. Because spectral CT extracts complete tissue characterization information from the imaged subject, many clinical applications can benefit from spectral CT technology, including improved material discrimination.

[0005] One challenge to more effectively use semiconductor-based PCDs for spectral CT is to perform material decomposition of projection data in a robust and efficient manner. For example, correction for pileup in the detection process can be imperfect, and these imperfections reduce the reliability of the material configurations resulting from material decomposition.

[0006] In photon-counting CT systems, semiconductor-based detectors using direct conversion are designed to resolve the energy of individual incident photons and generate measurements of multiple energy bin counts during each integration period. However, due to the detection physics of such semiconductor materials (e.g., CdTe / CZT), the detector's energy response is significantly degraded / distorted by charge sharing, k-escape, and scattering effects in the energy accumulation and charge induction process, as well as electronic noise in the associated front-end electronics. Due to the finite signal induction time, pulse pileup also distorts the energy response at high count rates.

[0007] Due to the heterogeneity and complexity of sensor materials in integrated detection systems, it is extremely difficult to accurately model such detector responses for photon-counting detectors solely based on physical theory or Monte Carlo simulations based on modeling of signal-inducing processes, which determines the accuracy of the forward model for each measurement. Furthermore, uncertainties in incident X-ray tube spectral modeling introduce additional errors into the forward model. All these factors ultimately lead to inaccurate material decomposition from PCD measurements and reduced spectral quality.

[0008] To address these issues, calibration methods have been proposed. The general idea is to use multiple transmission measurements with various known path lengths to calibrate a forward model so that it matches the calibration measurements. Some ideas have been applied to the estimation of X-ray spectra in conventional CT (see Non-Patent Documents 1 and 2) and then adopted to PCD measurements to estimate the spectral response of complex systems (see Non-Patent Document 3). However, there may be many variations in the detailed design and implementation of the calibration method, especially considering the feasibility of application in full third-generation CT geometries, which has not been demonstrated or documented in the literature to date. [Prior art documents] [Non-patent literature]

[0009] [Non-Patent Document 1] Emil idky et al., "A robust method of x-ray source spectrum estimation from transmission measurements: Demonstrated on computer simulated, scatter-free transmission data", Journal of Applied Physics, 2005, Vol.97, Issue.12, p124701 [Non-patent document 2] Duan et al., "CT scanner x-ray spectrum estimation from transmission measurements", Medical Physics, February 2011, Vol. 38, Issue 2, p993 [Non-patent document 3] Jannis Dickmann et al., "A count rate-dependent method for spectral distortion correction in photon counting CT", Proc. SPIE 10573, Medical Imaging 2018: Physics of Medical Imaging, March 9, 2018, p1057311 Summary of the Invention [Problem to be solved by the invention]

[0010] One of the problems to be solved by the embodiments disclosed in this specification and the drawings is to efficiently calibrate the detector response in a photon-counting detector having a plurality of groups each consisting of a plurality of subpixels. However, the problem to be solved by the embodiments disclosed in this specification and the drawings is not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0011] According to an embodiment, a photon-counting X-ray CT apparatus includes a photon-counting detector and a processing unit. The photon-counting detector includes a plurality of combined pixels each composed of a plurality of sub-pixels. The processing unit acquires calibration data based on the sub-pixels, generates combined calibration data according to a spatial resolution based on the calibration data, and calibrates a detector response of the photon-counting detector according to the spatial resolution based on the combined calibration data. [Brief explanation of the drawings]

[0012] [Figure 1]FIG. 1 shows an example of the configuration of a photon counting X-ray CT device. [Figure 2] FIG. 2 shows an example of a PCD bin response function Sb(E) for a photon counting detector, with each curve in FIG. 2 representing an example function for each energy bin. [Figure 3] FIG. 3 shows an example of a 3×3 sub-pixel pattern with a one-dimensional anti-scatter grid. [Figure 4] Figure 4 shows the workflow for material decomposition calibration and processing. [Figure 5] Figure 5 shows the normalized linear attenuation coefficients of different materials. [Figure 6] FIG. 6 shows a schematic diagram of the calibration structure design, where a pile-up correction table Pb is generated and used for each mA separately. [Figure 7] FIG. 7 shows a schematic diagram of another calibration structure design in which a universal pile-up correction table Pb is generated for the entire mA range. [Figure 8A] FIG. 8A shows an example of a summing scheme for decomposition calibration and processing. [Figure 8B] FIG. 8B shows an example of a summing scheme for decomposition calibration and processing. [Figure 8C] FIG. 8C shows an example of a summing scheme for decomposition calibration and processing. [Figure 8D] FIG. 8D shows an example of a summing scheme for decomposition calibration and processing. DETAILED DESCRIPTION OF THE INVENTION

[0013] In the following description, references to "one embodiment" or "an embodiment" mean that a particular feature, structure, material, or characteristic described in connection with that embodiment is included in at least one embodiment of the present application, but do not mean that it is present in all embodiments.

[0014] Thus, appearances of the phrases "in one embodiment" or "in an embodiment" below are not necessarily all referring to the same embodiment of the present application. Furthermore, the particular features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments.

[0015] In typical semiconductor-based PCDs, the detectors typically feature a pixelated design, using smaller subpixels in groups equivalent to traditional pixel sizes. That is, the PCDs have multiple groups of subpixels. This allows for high-resolution imaging with smaller pixel sizes, but different pixel patterns require different calibration schemes. This disclosure focuses on the pixelated pattern design of the detector and weighting methods for material decomposition calibration.

[0016] The following embodiment relates to a two-stage calibration method for a multicolor forward counting model for semiconductor-based PCDs to configure various pixel summing readout modes for imaging at different resolutions. The method involves: 1) calculating the flux-independent weighted bin response function S using the expectation-maximization (EM) method; wb (E) and 2) the pile-up correction term P b (E,N b ,N tot ) is composed of an estimate of S wb Once (E) is estimated from calibrations at multiple tube voltage (kVp) settings for each detector pixel, it is stored in the system as a software calibration table. Next, a pile-up correction term P for higher flux scans is calculated. b (E,N b ,N tot ) are used as input to estimate the path length of the reference material. Both tables are then used in material decomposition on the scan being run.

[0017] To correct for variations in detector response due to different PCD sub-pixel summation schemes, embodiments calibrate the forward model parameters based on various pixel readout modes.

[0018] Each subpixel (or combined pixel consisting of several subpixels) is calibrated individually following the same calibration and processing workflow. This applies to all air or subject scans used to calibrate the forward model and to patient / subject scans that use a different pixel readout scheme for imaging with the calibrated table.

[0019] A similar calibration is performed at each supported rotation speed to correct for differences in Anti-Scatter Grid (ASG) shadows for different sub-pixels or combined pixels at different rotation speeds.

[0020] 1 is a diagram showing an example of the configuration of a photon-counting X-ray CT apparatus 1. As shown in FIG.

[0021] The gantry 10 is a device that emits X-rays to the subject P (patient), detects the X-rays that pass through the subject P, and outputs them to the console 30.It includes an X-ray emission control circuit 11, an X-ray generator 12, a detector 13, a data acquisition circuit (Data Acquisition System: DAS) 14, a rotating frame 15, and a gantry drive circuit 16.

[0022] The rotating frame 15 is an annular frame that supports the X-ray generator 12 and the detector 13, which face each other across the subject P and are rotated at high speed in a circular orbit around the subject P by a gantry drive circuit 16.

[0023] The X-ray emission control circuit 11 is a device that functions as a high voltage generating unit and supplies a high voltage to the X-ray tube 12a, and the X-ray tube 12a generates X-rays using the high voltage supplied from the X-ray emission control circuit 11. Under the control of the scan control circuit 33, the X-ray emission control circuit 11 adjusts the tube voltage or tube current supplied to the X-ray tube 12a, thereby adjusting the amount of X-rays emitted to the subject P.

[0024] Furthermore, the X-ray emission control circuit 11 switches between the wedges 12b. Furthermore, the X-ray emission control circuit 11 adjusts the numerical aperture of the collimator 12c, thereby adjusting the X-ray emission range (fan angle or cone angle). There may also be cases where an operator manually switches between multiple types of wedges.

[0025] The X-ray generating device 12 is a device that generates X-rays and emits the generated X-rays to the subject P, and includes an X-ray tube 12a, a wedge 12b, and a collimator 12c.

[0026] The X-ray tube 12a is a vacuum tube that emits an X-ray beam toward the subject P using a high voltage supplied from the X-ray emission control circuit 11. As the rotating frame 15 rotates, the X-ray tube 12a generates an X-ray beam that spreads with a fan angle and a cone angle. For example, under the control of the X-ray emission control circuit 11, the X-ray tube 12a can continuously emit X-rays all around the subject P for complete reconstruction, or continuously emit X-rays for half reconstruction within an emission range (180° + fan angle) that allows half reconstruction. Furthermore, under the control of the X-ray emission control circuit 11, the X-ray tube 12a can intermittently emit X-rays (pulsed X-rays) at a previously set position (tube position). Furthermore, the X-ray emission control circuit 11 can change the intensity of the X-rays emitted from the X-ray tube 12a. For example, the X-ray emission control circuit 11 increases the intensity of X-rays emitted from the X-ray tube 12a at a specific tube position, and decreases the intensity of X-rays emitted from the tube 12a in areas other than the specific tube position.

[0027] The wedge 12b is an X-ray filter that adjusts the amount of X-rays emitted from the X-ray tube 12a. Specifically, the wedge 12b is a filter that transmits and attenuates X-rays emitted from the X-ray tube 12a so that the X-rays emitted from the X-ray tube 12a to the subject P have a predetermined distribution. For example, the wedge 12b is a filter obtained by processing aluminum so that it has a predetermined target angle or a predetermined thickness. Furthermore, the wedge is also called a wedge filter or a bowtie filter.

[0028] The collimator 12c is a slit that, under the control of the X-ray emission control circuit 11, narrows the irradiation range of the X-rays whose amount has been adjusted by the wedge 12b.

[0029] The gantry drive circuit 16 drives and rotates the rotating frame 15 so that the X-ray generator 12 and the detector 13 rotate in a circular orbit around the subject P.

[0030] Each time an X-ray photon enters, the detector 13 outputs a signal that can measure the energy value of the X-ray photon. This X-ray photon is, for example, an X-ray photon emitted from the X-ray tube 12a and passing through the subject P. The detector 13 includes a plurality of detection elements that output one pulse of an electrical signal (analog signal) each time an X-ray photon enters. The photon-counting X-ray CT device 1 counts the number of these electrical signals (pulses) to count the number of X-ray photons incident on each detection element. Furthermore, the photon-counting X-ray CT device 1 performs arithmetic processing on the signal to measure the energy value of the X-ray photon that outputs this signal.

[0031] The detection element includes, for example, a scintillator and an optical sensor such as a photomultiplier tube. In such a case, the detector 13 shown in FIG. 1 is an indirect conversion detector that uses a scintillator to convert incident X-ray photons into scintillator light and then converts the scintillator light into an electrical signal using an optical sensor such as a photomultiplier tube. The detection element may also be a semiconductor device such as cadmium telluride (CdTe) or cadmium zinc telluride (CdZnTe). In such a case, the detector 13 shown in FIG. 1 is a direct conversion detector that directly converts incident X-ray photons into an electrical signal.

[0032] For example, the detector 13 shown in FIG. 1 is a planar detector in which N detector elements are arranged in the channel direction (X-axis direction in FIG. 1) and M detector elements are arranged in the column direction (direction of the rotation center axis of the rotating frame 15 when the gantry 10 is not tilted, direction of the Z-axis in FIG. 1). The channel direction is also referred to as the column direction. The column direction is also referred to as the row direction. When a photon is incident, the detector element outputs one pulse of an electrical signal. The photon-counting X-ray CT device 1 identifies each pulse output from the detector element 131 and counts the number of X-ray photons that enter the detector element 131. Furthermore, the photon-counting X-ray CT device 1 performs arithmetic processing based on the intensity of the pulse and thereby measures the energy value of the counted X-ray photons.

[0033] The data acquisition circuitry 14 is a data acquisition system (DAS) that collects detection data on the X-rays detected by the detector 13. For example, the data acquisition circuitry 14 generates count data obtained by counting photons (X-ray photons) originating from X-rays that pass through the subject for each energy band, and transmits the generated count data to a console 30 (described later). For example, if X-rays are continuously emitted from the X-ray tube 12a while the rotating frame 15 is rotating, the data acquisition circuitry 14 collects a group of count data for the entire circumference (360 degrees). The data acquisition circuitry 14 can also collect data for each view. Furthermore, the data acquisition circuitry 14 transmits each count data acquired in relation to the position of the tube to the console 30 (described later). This tube position is information indicating the projection direction of the count data.

[0034] The couch device 20 is a device on which the subject P is placed, and includes a couch drive device 21 and a top board 22, as shown in FIG. 1 . The couch drive device 21 moves the top board 22 in the Z-axis direction to move the subject P into the rotating frame 15. The top board 22 is a board on which the subject P is placed. Furthermore, in this embodiment, a case will be described in which the relative position between the gantry 10 and the top board 22 is changed by controlling the top board 22. However, this is not a limitation of the embodiment. For example, if the gantry 10 is self-propelled, the relative position between the gantry 10 and the top board 22 can be changed by controlling the drive of the gantry 10.

[0035] Further, for example, the gantry 10 may perform a helical scan by rotating the rotating frame 15 while moving the top board 22, thereby scanning the object P in a spiral pattern. Alternatively, the gantry 10 may perform a conventional scan, i.e., scan the object P in a circular orbit by rotating the rotating frame 15 while keeping the position of the object P fixed after the top board 22 is moved. Alternatively, the gantry 10 may implement a step-and-shoot method, whereby conventional scans are performed in multiple scan regions by moving the position of the top board 22 at regular intervals.

[0036] The console 30 is a device that receives operations of the photon counting X-ray CT apparatus 1 from an operator and reconstructs X-ray CT image data using the projection data acquired by the gantry 10. As shown in FIG. 1 , the console 30 includes an input circuit 31, a display 32, a scan control circuit 33, a preprocessing circuit 34, a memory circuit 35, an image reconstruction circuit 36, and a processing circuit 37.

[0037] The input circuitry 31 includes a mouse, keyboard, trackball, switch, button, joystick, or the like, and is used by the operator of the photon counting X-ray CT apparatus 1 to input various commands or settings, and transfers information about these commands or settings received from the operator to the processing circuitry 37. For example, the input circuitry 31 receives from the operator acquisition conditions for X-ray CT image data, reconstruction conditions for reconstructing X-ray CT image data, image processing conditions for the X-ray CT image data, and the like.

[0038] The display 32 is a monitor viewed by the operator, and under the control of the processing circuitry 37, displays image data generated from X-ray CT image data to the operator, or displays a graphical user interface (GUI) for receiving various commands or settings from the operator via the input circuitry 31.

[0039] The scan control circuitry 33, under the control of the processing circuitry 37, controls the operation of the X-ray emission control circuitry 11, the gantry drive circuitry 16, the data acquisition circuitry 14, and the couch drive device 21, thereby controlling the data acquisition process by the gantry 10. For example, the scan control circuitry 33 sends sequence control commands to the data acquisition circuitry 14 to control the irradiation operation, as described in more detail below.

[0040] The pre-processing circuit 34 performs correction processes such as logarithmic conversion, offset correction, sensitivity correction, and beam hardening correction on the count data generated by the data acquisition circuit 14, thereby generating corrected projection data.

[0041] The memory circuitry 35 stores the projection data generated by the preprocessing circuitry 34. Furthermore, the memory circuitry 35 stores image data and the like generated by an image reconstruction circuitry 36, which will be described later. Furthermore, the memory circuitry 35 appropriately stores the processing results of a processing circuitry 37, which will be described later.

[0042] The image reconstruction circuit 36 ​​reconstructs X-ray CT image data using the projection data stored in the memory circuit 35. The reconstruction method here may include various methods, such as backprojection processing. Furthermore, the backprojection processing may include, for example, backprojection processing using a filtered backprojection (FBP) method. Alternatively, the image reconstruction circuit 36 ​​may also use an iterative approximation technique to reconstruct the X-ray CT image data. Furthermore, the image reconstruction circuit 36 ​​performs various types of image processing on the X-ray CT image data, thereby generating image data. The image reconstruction circuit 36 ​​then stores the reconstructed X-ray CT image data or various types of image data generated during the image processing in the memory circuit 35.

[0043] The processing circuitry 37 controls the operations of the gantry 10, the couch device 20, and the console 30, thereby performing overall control of the photon counting X-ray CT apparatus 1. Specifically, the processing circuitry 37 controls the scan control circuitry 33 to control the CT scan performed by the gantry 10. The processing circuitry 37 also generates calibration data (described below) and calibrates the detector response using the calibration data. Furthermore, the processing circuitry 37 controls the image reconstruction circuitry 36 to control the image reconstruction process or image generation process performed by the console 30. Furthermore, the processing circuitry 37 performs control so that various types of image data stored in the memory circuitry 35 are displayed on the display 32.

[0044] So far, we have explained the overall configuration of the photon-counting X-ray CT apparatus 1 according to the first embodiment. Here, the processing functions executed by each of the above circuits are stored in the form of programs executable by a computer in a storage circuit 35. Furthermore, each circuit reads and executes each program from the storage circuit 35, thereby performing the various functions described above.

[0045] In one example, programs corresponding to the operation of the data acquisition circuitry 14 are stored in the form of computer-executable programs in the memory circuitry 35. The processor 37 executes the programs in the data acquisition circuitry 14, sends instructions to and controls the data acquisition circuitry 14 to acquire data, and controls data transfer from the data acquisition circuitry 14. In a second example, the data acquisition circuitry 14 includes a processor that reads and executes each program from the memory circuitry 35 to perform the function corresponding to each program.

[0046] Furthermore, the term "processor" used in the above description refers to a circuit such as a central processing unit (CPU), a graphics processing unit (GPU), or an application specific integrated circuit (ASIC), or a programmable logic device (e.g., a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), or a field programmable gate array (FPGA)). The processor reads and executes a program stored in a memory circuit to perform a function. Furthermore, instead of storing the program in a memory circuit, the program may be directly installed in the processor's circuit. In this case, the processor reads and executes the program installed in the circuit to perform a function. Furthermore, with regard to the processor according to this embodiment, instead of each processor being configured as a single circuit, multiple independent circuits may be combined to form a single processor to perform a function.

[0047] In transmittance measurements using a photon counting energy-resolving detector (PCD), the forward model can be formulated as follows:

[0048]

number

[0049] In equation (1), S b (E) represents the bin response function defined by the following equation (2).

[0050]

number

[0051] In equation (2), R(E,E') is the detector response function, and E bL and E bH are the low and high energy thresholds for each counting bin. Figure 2 shows a typical S b (E) shows an example model, where the long tail above the energy window is induced by charge sharing, k-escape, and scattering effects. The low-energy tail is mainly due to the finite energy resolution from the associated electronic noise. N0 is the total flux from the air scan. μ m and l m are the linear attenuation coefficient and the optical path length of the mth reference material, respectively. w(E) is the normalized incident X-ray spectrum. In practice, w(E) and S b (E) are not known exactly, but they are expressed as a single term S wb (E)=w(E)S b (E), which is hereafter referred to as the weighted bin response function. wb If (E) can be calibrated by measurement, the decomposition problem under low flux conditions can be fully resolved. wb (E) is an example of the first parameter. That is, the first parameter is a parameter that depends on the bin response function Sb(E).

[0052] For high flux scan conditions (e.g., only a few percent of pulse pile-up), pulse pile-up introduces additional spectral distortions into the measurement. One way to correct for the pile-up effect is to introduce an additional correction term (e.g., Dickmann uses the measured count rate as input). This type of additional calibration also contributes to the flux-independent weighted bin response S wb (E) is based on an accurate estimate.

[0053] Under typical CT clinical scanning conditions, it is common for some measurements to have pulse pile-up of several percent or more. The resulting effect on material decomposition depends not only on the flux but also on the measured spectrum.

[0054] Semiconductor-based PCDs typically use a minimum pixel design, where the smallest readout units are grouped into various patterns for summed readout, allowing imaging at various resolutions. Different pixel summation schemes vary the detector response due to slightly different charge sharing and crosstalk effects caused by different combined pixel sizes and shapes resulting from the subpixel size and shape. Therefore, calibration and data processing must take this variation into account to obtain more accurate material decomposition results.

[0055] Third-generation CT systems often employ ASGs to eliminate scattered photons for cleaner measurements and better image quality. The small detector pixel design of photon-counting computed tomography (PCCT) theoretically eliminates the dead zones found in traditional scintillator-based detectors, allowing optimal performance to be achieved using the same or different ASG designs. However, when the same ASG design is used, the ASGs cast different shadows on each subpixel, resulting in even greater differences in charge sharing and crosstalk effects (see Figure 3). Therefore, this variable must be considered in the forward model calibration.

[0056] In particular, Figure 3 shows an example of a 3x3 subpixel pattern with a 1D ASG forming a combined pixel. However, in general, subpixel patterns of various sizes can be used, such as an nxm subpixel pattern or an nxn subpixel pattern, where n >= 2. For columns c1 and c3, the charge-sharing crosstalk effect is significantly reduced from the left / right due to the blocked area under the ASG plate, resulting in a different effective detector response than the center column c2.

[0057] The disclosure presented herein includes a two-stage calibration method for the PCD forward model of material decomposition, which consists of two parts: 1) the flux-independent weighted bin response function S using the expectation-maximization (EM) method; wb (E), and 2) the energy (E) and the number of measured bins (N b ,N tot ) pile-up correction term P b (E,N b ,N tot ) where N b is the number of individual bins, N tot is the total number of all energy bins. The calibrated forward model can be expressed as follows:

[0058]

number

[0059] A series of slab measurements using known materials and path lengths are used to calibrate the forward model described above.

[0060] Here, instead of using only two substances as in prior art (see, e.g., Dickmann), this method uses 2–5 different substances, such as polypropylene, water, aluminum, titanium / copper, and k-edge materials, to measure the weighted bin response function S at low fluxes. wb (E) Calibrate. More selective materials used in the calibration reduce the number of total path lengths and give comparable or better results.

[0061] Step 1: Using appropriate tube spectral modeling to capture the characteristic peaks of the incident spectrum and a physical model to simulate the spectral response of the photon-counting detector, S wb An initial estimate of (E) can be generated using the EM method (see, e.g., Sidky). wb (E) can be reliably estimated for this very ill-conditioned problem based on several transmittance measurements.

[0062] where b(E,N b ,N tot ) is assumed to be constant in step 1. The calibrated forward model can be simplified to a system of linear equations shown in equation (4).

[0063]

number

[0064] Typically, the number of data measurements (M) is proportional to the number of unknowns (E max ), which is much less than the eigenvalues. Assuming a Poisson distribution of the data collection, we derive an iterative EM algorithm to find the unknown energy bin response function S, as described below. wb The best estimate of (E) can be found.

[0065] When low-flux data collection is used to estimate bin response functions, a pile-up effect correction P b is assumed to be a known term (e.g., a constant). Therefore, the model is simplified to the following equation (5):

[0066]

number

[0067] Here, equation (6) below represents the attenuated path length for the jth measurement.

[0068]

number

[0069] Therefore, each measurement value j can be expressed as in the following equation (7).

[0070]

number

[0071] For M measurements, the data collection can be written in matrix form as follows:

[0072]

number

[0073] Or A·S wb =N b The following equation holds true.

[0074] By applying the EM iterative algorithm, S wb can be estimated as follows:

[0075]

number

[0076]

number

[0077] S wb The update equation for (E) is given by the following equation (10).

[0078]

number

[0079] Step 2: Calibration of each detector pixel at each tube voltage (kVp) setting to S wb Once (E) is estimated, it is stored in the system as a software calibration table, which is used to calculate the pile-up correction term P for high flux scans. b (E,N b ,N tot ) are used as inputs to further estimate the path length of the reference material. Both tables are then used in material decomposition of the object / patient scan to estimate the path length of the reference material.

[0080] The calibration table is updated from time to time based on variations in system / detector performance. This can also be designed as an iterative procedure. If the image quality on the quality check phantom is not sufficient, the calibration process is repeated using the updated calibration table from the last iteration as an initial guess.

[0081] A high-level workflow of the above process is shown in Figure 4. Steps 1)-4) represent the calibration workflow, and steps 5)-8) show how the calibration table is used in an operational scan of a patient / subject to generate a spectral image.

[0082] First, a series of low-flux scans of various material slabs are collected at each tube kVp setting, which is the peak potential applied to the X-ray tube. A typical CT system supports several kVp settings, from 70 to 140 kVp, to generate different energy spectra from the X-ray tube for different scan protocols. For CT scans, both mA and kVp must be preselected before turning on the tube. Next, a low-flux weighted bin response function, S, is calculated. wb is estimated, and this estimated S wb and a high flux slab scan was used to calculate the pile-up correction term P b Additional parameters are estimated for each detector pixel, S wb and P b Using the estimated calibration table, the quality of the calibration is checked on a quality phantom, e.g., a homogeneous water phantom or a phantom with multiple inserts of homogeneous known material. The image quality is assessed using predefined standards and, if passed, the current calibration table is saved and used for the next patient / subject scan data processing. Otherwise, the procedure continues. wb and P bUse the last iteration as the initial estimate and repeat the first three steps. Here, the generally considered criteria are the accuracy, uniformity, spatial resolution, noise, and artifacts of the image CT number. To check the quality of this calibration, it is necessary to check all these metrics, especially the accuracy and artifacts such as rings and bands in the image indicating insufficient calibration.

[0083] To select the optimal material and path length for this calibration, a normalized linear attenuation coefficient vs. energy curve can be used (see Figure 5). For selecting different ones from each other, for example, polypropylene, water, aluminum, and titanium are a good group of suitable combinations for such calibrations covering a wide range of common materials present in the human body.

[0084] To meet the low-flux conditions through calibration measurements to minimize the pile-up effect in the flow diagram, in Step 1, it can be selected to use "nτ < x". Here, x is about 0.005 - 0.01, n is the pixel counting rate at the lowest dose setting in the X-ray tube, and τ is the effective dead time of the PCD application-specific integrated circuit (ASIC). By meeting this condition, the shortest path length of each selected calibration material can be calculated, and the remaining path lengths can be selected either by equal spacing in the path length or the obtained measurement counting rate.

[0085] For the calibration of the pile-up correction term P in Step 3 b the same slab material and path length are used for the scan at a high mA setting. The calibration data are grouped by mA to generate different correction tables for each mA setting (see Figure 6), or a universal correction table for continuous mA settings is generated, including measurements in all flux ranges (e.g., low - high mA, high - low mA, or the most frequently used value first) (see Figure 7).

[0086] The calibration measurements should be performed with sufficient statistics to minimize the effects of statistical fluctuations. One non-limiting example is to use statistics greater than 1000 times the typical integration period for the calibration data set to minimize statistical errors transferred in the calibration. Each energy bin b in the calibration measurements has a corresponding S wb (E) and P b (E,N b ,N tot ) is used to update the

[0087] The estimation is very ill-conditioned because only a limited number of measurements can be performed in a small number of energy bins. In this case, a good initial guess is essential for accurate estimation, as it provides additional constraints on the EM method. One design variation to accommodate non-ideal detectors is to use S with slight variations in the actual energy threshold settings, especially in the ASIC. b The aim is to allow a more flexible energy window for each bin in the initial guess of S. By setting the low threshold xkeV low and the high threshold ykeV high, the initial S b is expressed as the following equation (11).

[0088]

number

[0089] In equation (11), x and y can be chosen between 5 and 10 keV, allowing for some variation in ASIC characteristics while providing additional constraints on EM issues.

[0090] This calibration is performed pixel-by-pixel with each bowtie / filter configuration to capture spectral variations across the fan beam after the bowtie filter and detector response variations across different detector pixels.

[0091] The design described in this embodiment uses more than two materials in the calibration, which provides greater sensitivity to constraining the weighted bin response function estimation problem of photon counting detectors.

[0092] In addition, this method uses a high flux pile-up correction term, P b (Now E,N b ,N tot We use different parameterizations for the total count term N tot is introduced to better approximate the true pile-up phenomenon, which can significantly improve the model capability at higher flux conditions with fewer parameters.

[0093] Additionally, a range of different calibration path lengths can be used at different fan angles to improve the accuracy and efficiency of the calibration. The slab scan used to calibrate the forward model can be selected based on the imaging task that produces the best image quality.

[0094] Finally, an alternative scheme is presented for computing initial estimates of the weighted bin response functions by expanding the energy threshold window to accommodate non-ideal detector / ASIC performance.

[0095] To accommodate the various detector responses of different summing schemes using ASG, this embodiment presents separate calibrations of the forward model using different sub-pixel summing schemes. The calibration tables for each pixel configuration are stored in digital data format and appropriately used in data processing to reconstruct images at various resolutions.

[0096] Steps 1)-4) in Figure 4 are repeated for different pixel readout summation modes. A calibration table is saved for each mode and applied to scans of patients / subjects using the same readout configuration.

[0097] In semiconductor-based photon-counting detectors for CT, a pixel size of 200–500 μm is typically chosen for optimal performance. This provides a good balance between charge-sharing effects and pulse pile-up under CT scan flux conditions. This is smaller than conventional scintillator-based pixels, which are typically about 1 × 1 mm. Therefore, image processing readout can have multiple modes, combining different numbers of subpixels into one or more combined pixels.

[0098] Combined pixel readout mode (N r ×N c ) (i.e., combined multiple sub-pixel readout mode), the forward model calibration can be based on the sum (or average) measurements of the combined pixels.

[0099]

number

[0100]

number

[0101] Figure 8 shows an example of a 3x3 subpixel scheme that forms a combined pixel equivalent to a conventional EID pixel. Four different summation schemes can be used to read out the data: for example, "3x3", "1x3", "3x1", and "1x1", which represent four possible resolutions for the resulting image.

[0102] For example, combined calibration data may be generated according to spatial resolution. Such combined calibration data may be generated, for example, by performing each of the scans described above to obtain counts for each energy bin and combining the resulting counts for a subpixel pattern. Such combined calibration data may effectively calibrate the detector response of the PCD according to spatial resolution.

[0103] The disclosed calibration scheme for different pixel summation patterns for combined pixels is not limited to a particular sub-pixel pattern or a particular forward model, but can be applied to any forward model calibration based on measurements of known materials at different path length samples. Furthermore, the counts obtained for the sub-pixel pattern can be taken to obtain an average over the sub-pixel pattern, and a normalization factor is applied to the average count based on an estimate of the anti-scatter grid (ASG) shadow on the sub-pixels.

[0104] FIG. 8 illustrates various summation schemes for resolution calibration and processing. In particular, in A), summation is performed on combined pixels, e.g., in a "3x3" summation mode. That is, in combining counts for a subpixel pattern, summation is performed across the entire subpixel pattern. In B), summation is performed column-wise, e.g., in a "1x3" summation mode. That is, in combining counts for a subpixel pattern, summation is performed across the columns of the subpixel pattern. In C), summation is performed channel-wise, e.g., in a "3x1" summation mode. That is, in combining counts for a subpixel pattern, summation is performed across the channels of the subpixel pattern. Finally, in D), calibration is based on individual subpixels. Subject scan material decomposition can select one of the summation patterns with a corresponding calibrated table. For example, during calibration data acquisition, the above-described combined calibration data is generated and stored in a table. That is, when subpixel-based calibration data is acquired, the combined calibration data is generated and stored in a table. Alternatively, after collecting the calibration data, the combined calibration data described above is generated and stored in a table before a subject scan is performed. Then, when a subject scan is performed, the detector response of the photon counting detector is calibrated based on the stored table. This reduces the amount of processing during a subject scan, thereby reducing the load on the processing system or increasing the processing speed.

[0105] The combined calibration data may be generated at the time of subject scanning, or may be generated when the combined calibration data is used, i.e., when calibrating the detector response of the photon-counting detector. This allows for less data to be stored between the time the calibration data is collected and the time the subject scan is performed, thereby saving memory space.

[0106] By calibrating each subpixel and averaging the material decomposition results in the estimated path length sinograms for various combined readout modes, the disclosed method directly addresses the averaging effect in the calibration process and avoids the complexity of weighting the data after the decomposition step.

[0107] To reduce calibration time, all slab calibration data can be acquired using the smallest detector pixel units (subpixels, also known as micropixels) and then digitally combined into various summation patterns by data processing, which can be performed in the front-end electronics or detector / system firmware (FPGA).

[0108] Many modifications and variations of the embodiments presented herein are possible in light of the above teachings. It is therefore to be understood that, within the scope of the appended claims, the present disclosure may be practiced otherwise than as specifically described herein.

[0109] At least one of the embodiments described above can efficiently calibrate the detector response in a photon-counting detector having multiple groups of multiple sub-pixels.

[0110] Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, modifications, and combinations of embodiments can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0111] 1. Photon-counting X-ray CT equipment 30 Console 37 Processing circuit

Claims

1. 1. A photon-counting X-ray CT apparatus capable of using a plurality of different summation schemes to read out data for sub-pixel patterns of a photon-counting detector having a plurality of groups each consisting of a plurality of sub-pixels, thereby providing a plurality of spatial resolutions for the resulting image, comprising: the photon-counting detector; obtaining calibration data based on the sub-pixels; generating combined calibration data according to the spatial resolution based on the calibration data; a processor for calibrating a detector response of the photon-counting detector according to the spatial resolution based on the combined calibration data; Equipped with the processing unit performs a plurality of low flux scans, including an air scan and a scan using a plurality of slabs of different materials, in the sub-pixel pattern, at a plurality of initial current intensities and tube voltage settings of an x-ray tube to obtain counts for each energy bin, and combines the obtained counts for the sub-pixel pattern to generate the combined calibration data; A photon-counting X-ray CT system that performs reference material discrimination in a subject scan and estimates the reference material based on the combined calibration data.

2. 2. The photon-counting X-ray CT apparatus of claim 1, wherein combining the obtained counts for the sub-pixel pattern comprises summation performed over the entire sub-pixel pattern.

3. 2. The photon counting X-ray CT apparatus of claim 1, wherein combining the obtained counts for the sub-pixel pattern comprises summation performed across a column or channel direction of the sub-pixel pattern.

4. 2. The photon-counting X-ray CT apparatus of claim 1, wherein combining the obtained counts for the sub-pixel pattern comprises averaging over a sub-pixel pattern and applying a normalization factor to the averaged counts based on an estimate of an anti-scatter grid shadow on the sub-pixel pattern.

5. 2. The photon-counting X-ray CT apparatus of claim 1, wherein combining the obtained counts for the sub-pixel pattern comprises averaging over a column or channel direction of the sub-pixel pattern and applying a normalization factor to the averaged counts based on an estimate of an anti-scatter grid shadow on the sub-pixels.

6. 6. The photon counting X-ray CT apparatus according to claim 1, wherein the processing unit further estimates a first parameter that depends on energy based on the combined counts for the sub-pixel pattern.

7. 7. The photon counting X-ray CT apparatus according to claim 6, wherein the processing unit further estimates a second parameter that depends on a total count number of all energy bins based on the estimated first parameter.

8. 8. The photon-counting X-ray CT apparatus according to claim 7, wherein the processing unit repeats, using each of the plurality of slabs, estimating the first parameter and estimating the second parameter at a different current intensity other than the initial current intensity and at the same tube voltage, to acquire the second parameter for different current intensities.

9. 9. The photon counting X-ray CT apparatus according to claim 8, wherein the processing unit performs an initial estimate of the first parameter based on a detector response function and low and high energy thresholds for each counting bin.

10. The photon counting X-ray CT apparatus according to any one of claims 6 to 9, wherein the first parameter depends on a bin response function.

11. 10. The photon counting X-ray CT apparatus according to claim 7, wherein the second parameter is related to a pile-up correction term.

12. the processing unit repeatedly obtains the counts for each energy bin and combines the counts at different current intensities other than the initial current intensity and at the same tube voltage using each of the plurality of slabs, and obtains a universal table of estimated second parameters that depend on the total counts of all energy bins based on the first parameters over an entire current intensity range; 7. The photon counting X-ray CT apparatus of claim 6, further comprising: performing reference material decomposition in a subject scan to estimate the reference material based on the first parameter and the second parameter for combined counts for the sub-pixel pattern.

13. 13. The photon counting X-ray CT apparatus according to claim 8, wherein the processing unit evaluates image quality of a projection image by performing a phantom scan using the first parameter and the second parameter, and if the image quality of the projection image satisfies a predefined standard, estimates the reference material by using the first parameter and the second parameter for the reference material discrimination in a subject scan.

14. The photon-counting X-ray CT device according to any one of claims 1 to 13, wherein the processing unit generates the combined calibration data and stores it in a table when acquiring calibration data based on the sub-pixels or before a subject scan, and calibrates the detector response of the photon-counting detector based on the table when the subject scan is performed.

15. The photon counting X-ray CT device according to any one of claims 1 to 13, wherein the processing unit generates the combined calibration data during a subject scan or when calibrating a detector response of the photon counting detector.

16. A method performed by a photon-counting X-ray CT device capable of using a plurality of different summation schemes to read out data for subpixel patterns of a photon-counting detector having a plurality of groups of a plurality of subpixels, and capable of providing a plurality of spatial resolutions for the resulting image, comprising: a processing unit provided in the photon counting X-ray CT apparatus, obtaining calibration data based on a plurality of sub-pixels constituting each of a plurality of groups in the photon-counting detector; generating combined calibration data according to the spatial resolution based on the calibration data; calibrating a detector response of the photon-counting detector according to the spatial resolution based on the combined calibration data; A method comprising: The processing unit performing a plurality of low flux scans, including an air scan and scans using a plurality of slabs of different materials, at a plurality of initial current intensities and tube voltage settings of an x-ray tube for each scan, in the sub-pixel pattern to obtain counts for each energy bin, and combining the resulting counts for the sub-pixel pattern to generate the combined calibration data; A method for performing reference material discrimination on a subject scan performed on the photon counting X-ray CT device and estimating reference materials based on the combined calibration data.

Images